For TS I had wanted a way for the monitor routines to know that they were
in the final iteration (time-step) so I put in a hack with a time step == -1.
Lisandro proposed and implemented a far better solution of just making sure
that ts->convergedreason was set for EXACTLY the last call to monitor and using
that to indicate the final time-step.
I suggest we do the same thing for KSP and SNES. This will require moving
around slightly the location of Monitor() in the solvers but I took a quick
look at the codes and it looks doable.
Does this sound like a reasonable change to the code?
Barry
